Joshua Williams
Joshua Williams
Sometimes we want to end before Llama decides to generate `[end of text]`, which is the reason this feature was requested. But, sometimes we also might want multiple stop conditions....
Closing in favor of https://github.com/ggerganov/llama.cpp/pull/769
Issues should stay in the "New Issues" pipeline until they've been estimated (and don't just do this yourself - multiple members are supposed to estimate every task)
Issues should stay in the "New Issues" pipeline until they've been estimated (and don't just do this yourself - multiple members are supposed to estimate every task)
Issues should stay in the "New Issues" pipeline until they've been estimated (and don't just do this yourself - multiple members are supposed to estimate every task)
@a-bainbridge I know you've done some work on this. If this issue is redundant, say so and I'll remove it. Otherwise, just let me know what progress has been made...
Great, I thought that was the case. I just directly transcribed my issues list from before the start of the sprint, so I'm not surprised there's overlap.
Moved to backlog
My mistake, that was an accident. Moving it back
This issue specifically is a forceful reboot - the safety manager needs to skip talking to the node entirely and talk to the OS. So, it is not a recovery...